Leishmania Promastigotes Lack Phosphatidylserine but Bind Annexin V upon Permeabilization or Miltefosine Treatment

The protozoan parasite Leishmania is an intracellular pathogen infecting and replicating inside vertebrate host macrophages. A recent model suggests that promastigote and amastigote forms of the parasite mimic mammalian apoptotic cells by exposing phosphatidylserine (PS) at the cell surface to trigger their phagocytic uptake into host macrophages. PS presentation at the cell surface is typically analyzed using fluorescence-labeled annexin V. Here we show that Leishmania promastigotes can be stained by fluorescence-labeled annexin V upon permeabilization or miltefosine treatment. However, combined lipid analysis by thin-layer chromatography, mass spectrometry and 31P nuclear magnetic resonance (NMR) spectroscopy revealed that Leishmania promastigotes lack any detectable amount of PS. Instead, we identified several other phospholipid classes such phosphatidic acid, phosphatidylethanolamine; phosphatidylglycerol and phosphatidylinositol as candidate lipids enabling annexin V staining.


Introduction
Lipids are essential for the structural and functional integrity of cells. As the predominant constituents of cellular membranes, lipids compartmentalize cellular functions and are involved in various aspects of signal transduction. One major class of lipids in eukaryotic cell membranes is represented by phospholipids, consisting of a glycerol backbone, two fatty acyl residues, and a polar head group at the sn-3 position. The polar head group consists of a phosphate residue which is (except for phosphatidic acid, PA) esterified by an alcohol such as choline (to form phosphatidylcholine, PC) or ethanolamine (phosphatidylethanolamine, PE), the amino acid serine (phosphatidylserine, PS) or the carbohydrate inositol (phosphatidylinositol, PI). Among all these phospholipids, PS is a relatively minor constituent of most biological membranes. However, the low abundance of PS is outweighed by its physiological importance. Under normal conditions, PS is restricted to the inner plasma membrane leaflet in eukaryotic cells [1]. Any change in this distribution generally triggers a physiological event such as the clearance of apoptotic cells or the internalization of viruses by host cells [2,3,4]. PS has also been implicated in the infectivity of Leishmania, an obligate, intracellular parasite of humans and other mammals that infect cells of the mononuclear phagocyte lineage. The parasite has a digenic life cycle, residing as flagellated extracellular promastigote in the gut of the insect vector and as obligatory intracellular aflagellated amastigote found in the parasitophorous vacuoles of mammalian macrophages. A critical point in this host-parasite interaction involves the attachment to and invasion of host macrophages, initially by the promastigotes and subsequently by amastigotes. Both promastigotes and amastigotes use receptormediated phagocytosis for invasion. Furthermore, there is evidence that exposure of PS on the cell surface of the parasite mimics apoptosis and encourages the macrophages in the host organism to phagocytose the parasite [5,6,7,8]. Notably, in these studies PS exposure has been detected by flow cytometry based on reactivity with either anti-PS antibodies or annexin V. However, despite being used extensively to detect externalization of PS, neither anti-PS antibodies nor annexin V are specific for this lipid and also bind other phospholipids such as phosphatidylglycerol (PG) and phosphatidylinositol-4,5-bisphosphate [9]. Thus, a direct proof for PS exposure by Leishmania parasites is currently lacking. Even the presence of PS in Leishmania has not been firmly established. Previous studies of Leishmania lipid compositions by thin layer chromatography-based methods have reported the presence of PS in several Leishmania species [10,11,12], while other studies based on mass spectrometry analysis failed to detect this lipid [13,14,15]. Here, we performed a combined analysis of Leishmania phospholipid classes and their ability to bind annexin V. Our findings show that upon permeabilization or miltefosine treatment in vitro grown Leishmania promastigotes are able to bind annexin V but lack any detectable amount of PS. Instead, we identified several other phospholipid classes as candidate lipids enabling annexin V staining.

Annexin V Binding of L. donovani Promastigotes
To investigate whether L. donovani promastigotes can bind annexin V, we first permeabilized the parasites in the presence of 2.5 mM Ca 2+ and 125 ng/mL annexin V-FITC by electroporation. This treatment resulted in strong FITC labeling of the parasites; in some cases, however, this labeling was restricted to inner structures (Supplementary Figure S1). By contrast, untreated parasites did not show a significant FITC labeling and, thus, binding of annexin V. Furthermore, we incubated parasites with miltefosine, a potent anti-leishmanial drug inducing an apoptosis like death [16], which resulted in annexin V-FITC and propidium iodide positive staining of the parasites (Supplementary Figure S1). Since annexin V preferentially interacts with membranes containing PS, we next analyzed total lipid extracts from untreated and miltefosin-treated L. donovani promastigotes for the presence of PS by thin-layer chromatography (TLC). Consistent with our previous results [17] we could neither detect significant concentrations of PS in the lipid extracts from untreated nor from miltefosine-treated parasites by this method.

L. donovani Promastigotes Lack Phosphatidylserine
To corroborate that L. donovani promastigotes lack detectable levels of PS, total lipids were extracted from the parasites and fractionated by reversed phase HPLC coupled to electrospray ionization tandem MS using collision-induced dissociation. Major phospholipid species could be identified as: PC (mainly diacyl species), PE (diacyl and alkenylacyl (plasmalogen) species), PI (diacyl and alkenylacyl species), IPC and cardiolipins ( Figure 1; Table 1). No hydrolysis of the plasmalogen species [18] due to the slight acidic conditions during the chromatographic separation was detected. We searched for PS species by scanning the MS 2 spectra for the characteristic neutral loss of 87.0 amu (C 3 H 5 NO 2 ) but no significant levels of PS could be detected by this highly sensitive MS method. To prove that low abundant PS species can be detected in biological extracts employing the described HPLC-MS method, 0.1% PS (18:1/18:1) was added to a phospholipid extract of Escherichia coli. In the HPLC-MS analysis of the spiked phospholipid extract of E. coli PS (18:1/18:1) was easily detectable even though it co-eluted with abundant PG species (Supplementary Figure S2). These PG species proved to give the highest ion yield and were thus eminently detectable in equimolar mixtures of six lipid standards containing PS (16:0/18:1), PE (16:0/18:1), PC (16:0/18:1), PG (16:0/18:1), PA (16:0/18:1) and cardiolipin (4618:1) (Figures S3, S4).
Phospholipid Analysis in L. donovani Promastigotes by MALDI-TOF Mass Spectrometry and 31

P NMR Spectroscopy
As an alternative analytic method to detect PS, total lipid extracts were subjected to matrix-assisted laser desorption and ionization time-of-flight (MALDI-TOF) mass spectrometry. This method is known to be affected only very moderately by sample impurities that might be present in the investigated extracts [19]. Only the mass regions where PS peaks could be expected are shown. The positive ion spectrum was found to contain mainly PC species (Figure 2, upper panel). This is not surprising because the permanent positive charge of the quaternary ammonium renders PC the highest detectability. This leads to the suppression of less sensitively detectable phospholipid species [20]. Such suppression effects are well known in the context of soft-ionization MS methods that generate ''quasimolecular'' ions. This is a particular problem regarding PE: as this phospholipid is zwitterionic, it is suppressed by PC in the positive ion mode and by more acidic phospholipids such as PS or PI in the negative ion mode [19]. In order to overcome potential suppression effects in the positive ion mode, an alkaline matrix (9-AA) that is more suitable than 2,5dihydroxybenzoic acid (DHB) to record negative ion MALDI mass spectra [21] was also used. Negative ion MS detected PE, IPC and PI species (Figure 2, lower panel), while not even traces of PS were detectable. A detailed assignment of all detected peaks is provided in Table 2. No major efforts were undertaken to assign the fatty acyl compositions of IPC or to analyze cardiolipin that is rather difficult to detect by MALDI-TOF MS due to its higher mass in comparison to standard phospholipids [22,23]. In addition to the analysis of the total extract, combined TLC/negative ion MALDI MS was also attempted in order to exclude effects of ion suppression: although PS was detectable in artificial lipid mixtures and after spiking the cellular extract with PS, no PS was detectable in the native extract. Thus, MALDI MS confirmed the ESI MS results (vide supra) and there is obviously no PS in Leishmania promastogotes.
Finally, lipid extracts were also analyzed by 31 P NMR spectroscopy. This experimental approach provides the quantitative determination of all major phospholipid classes without the need of major sample work-up or separation into the individual lipid classes. The lipid mixture of interest is simply ''solubilized'' in a suitable detergent system in order to suppress the aggregation of phospholipids that would result in severe line-broadening and would make the differentiation of the individual phospholipid classes impossible [24]. However, we failed to detect any PS while the other phospholipid classes were easily detectable (Supplementary Figure S5). Although NMR suffers from much lower sensitivity in comparison to MS and should be cautiously interpreted, these NMR data are in line with the MS data (vide supra).

Lack of PS is a General Feature of Leishmania Promastigotes
To investigate whether Leishmania parasites lack PS in general, total lipid extracts were prepared from L. major, L. mexicana, L. amazonensis, L. tropica, L. guyanensis and L. shawi promastigotes and analyzed by TLC-MALDI imaging [25]. None of the investigated parasite strains showed any detectable amounts of PS (data not shown), although it is normally detectable in amounts of about 30 pg [26]. Further analysis of the lipid extracts by HPLC coupled electrospray ionization tandem MS did also not reveal the presence of PS in any of the tested strains (Supplementary Table  S1 and data not shown). Finally, we metabolically labeled the various parasite species with [ 3 H]serine and subsequently analyzed for its incorporation into total lipids by TLC. In all parasite species, serine was abundantly incorporated into both glycerophospholipids (PE and PC) and sphingolipids (ceramides and IPC) but no radiolabel signal was obtain for PS ( Figure 3 and data not shown). We conclude that Leishmania promastigotes do not synthesize PS.

Identification of Annexin V Binding Lipid Species in L. donovani Promastigotes
To identify Leishmania lipids that bind annexin V, lipids were extracted from L. donovani promastigotes and separated by TLC using chloroform/methanol/water (65/25/4, v/v/v) as solvent system ( Figure 4A). From ten individual TLC spots, lipids were extracted and analyzed by overlay assays [27]. To this end, lipids were spotted onto nitrocellulose and incubated with annexin V-FITC in the presence and absence of Ca 2+ . Under these conditions, Leishmania lipids extracted from regions 3 (PC, PI), 5 (cardiolipin), and 6 (PE) were found to bind to annexin V-FITC when Ca 2+ was present ( Figure 4B). Next, lipids from TLC region 3 were subjected to a second TLC separation using an alkaline solvent system to separate PI and PC ( Figure 4C). Upon extraction and re-protonation, both Leishmania lipids were examined for binding by annexin V-FITC in overlay assays. Ca 2+ -dependent binding was detected for PI but not for PC lipids ( Figure 4C). Control experiments with several standard lipids showed binding of annexin V-FITC to PS, PG, PE, and PI in the presence of Ca 2+ , while binding to PC, PA, SM, and cholesterol was negligible under these conditions ( Figure 4B).
To test whether the individual lipid classes identified in the overlay assays also promote annexin V-FITC binding when incorporated in lipid membranes, giant vesicles of different lipid compositions were generated and incubated with annexin V-FITC in the presence or absence of Ca 2+ ions. Fluorescence microscopy revealed Ca 2+ -dependent binding of annexin V-FITC to membranes containing PG, PE, or PI but not to pure PC vesicles ( Figure 5; Supplementary Figure S6). Annexin V conjugated with Alexa 568 as fluorophore gave equivalent results (Supplementary Figure S7). Quantitative assessment of annexin V-FITC binding to giant vesicles by flow cytometry revealed the strongest labelling for vesicles containing PG and PI. Labelling of vesicles containing PE was moderate while pure PC vesicles hardly exhibited fluorescence above the background contribution of the buffer (insets in uppermost flow cytometry graphs). Collectively, these data show that binding of annexin V is not restricted to PS but includes several other phospholipids such as PE, and particularly the acidic PI and cardiolipin that are present in Leishmania parasites in considerable amounts.

Discussion
In this study, we performed a combined lipid analysis of Leishmania promastigotes using three different techniques: thinlayer chromatography, (ESI and MALDI) MS and phosphorus-31 nuclear magnetic resonance spectroscopy. Based on these three complementary approaches, PC, PE, PI, PG, cardiolipin and inositol phosphorylceramide phospholipids were clearly detectable but there was no evidence for the presence of PS in lipid extracts derived from several in vitro grown Leishmania species, suggesting that the parasite does not synthesize PS under these conditions. Consistent with this notion, 3 H-serine labelling experiments further validated the absence of PS in promastigotes of several in vitro grown Leishmania species.
These findings are in line with several other MS analyses that found no evidence for the presence of PS in lipid extracts of L. donovani [15] and L. major promastigotes [13,14] but contrast with a recent MS study reporting the presence of PS in L. donovani promastigotes [28]. A potential reason for this discrepancy could be that in certain growth states/conditions such as late logarithmic phase, the parasites do synthesize PS (see below). The parasites may also be able to acquire PS from foetal calf serum that contains several phospholipids including PS [29] and is commonly used as a supplement to the culture media.
Beside L. donovani and L. major, promastigotes of L. mexicana, L. amazonensis, L. tropica, L. guyanensis, and L. shawi did not contain any detectable levels of PS neither in TLC-MALDI ''imaging'' nor in HPLC coupled electrospray ionization tandem mass spectrometry. Nevertheless, several of these parasite species have been reported to bind annexin V under certain conditions [16,30,31,32]. In the present report, we observed strong staining of L. donovani promastigotes by annexin V upon electroporation or miltefosine treatment despite the absence of detectable levels of PS. These observations call for caution regarding the use of fluorescencelabeled annexin V as the only approach to probe the presence of PS because our results imply that the parasites expose other lipids than PS that enable annexin V binding.
Using lipid overlay assays and giant vesicle membranes we found that annexin V binds not only to PS but also to several other phospholipids including PE and PI. In addition, annexin V has been reported to bind PA and phosphatidylinositol-4,5-bisphosphate [9]. Similar to PS, all these phospholipid species are enriched to varying extents on the cytoplasmic side of eukaryotic membranes, while sphingolipids (i.e. sphingomyelin and glycosphingolipids) are enriched in the exoplasmic leaflet [33]. During apoptosis, however, this asymmetric lipid arrangement in the plasma membrane is lost, resulting in drastic changes in the phospholipid composition of both leaflets [34,35,36,37]. Binding of annexin V to the cell surface of Leishmania parasites is therefore likely to be a consequence of changes in the plasma membrane lipid arrangement.
The lack of significant amounts of PS in Leishmania promastigotes implies that the parasite cannot take advantage of surface exposed PS during the initial infection process and calls into question the concept of PS-based apoptotic mimicry. In fact, this concept has not been firmly established for Leishmania parasites and is essentially based on the labeling of parasite (sub)populations with either anti-PS antibodies or annexin V. Our data suggest that the parasite may rearrange the plasma membrane distribution of other phospholipids such as PI, PE, PG and/or PA. Resting red cells, however, do not bind Annexin V in the presence of low calcium concentrations (i.e. 1 to 2 mM) although their outer membrane leaflet contains ,20% of PE (accounting for ,11 mol% of all phospholipids in this monolayer) and ,20% of PI (accounting for ,1 mol% of all phospholipids in this monolayer) [38]. Plasma membranes of L. donovani were found to contain an amount of PE, PI and PG that represented about 35, 4 and 2% of the total phospholipids [39]. Taken into account that the extent to which annexin V binds to a membrane is a complex function related to the membrane lipid composition, the free annexin V concentration and the calcium concentration [40], it is likely that global changes in the transbilayer arrangement of PE, PI, PG and/or PA trigger annexin V staining of the parasites. In support of this notion, we recently found that parasites lacking a plasma membrane lipid transporter but not wild-type parasites are strongly labeled by PE-binding peptides [17]. It remains to be evaluated whether PE, PI, PG and/or PA do contribute to the infectivity of parasites.
The genomes of Leishmania and other trypanosomatids encode two genes showing similarity to eukaryotic phosphatidyl serine/ ethanolamine base exchange enzymes [14]. This base exchange/decarboxylase cycle could in principle function to provide the parasites with PE rather than PS, thus accounting for its presence in the genome in the absence of detectable amounts of PS. However, studies by Zhang et al. [14,41,42] revealed that this pathway does not contribute to PE synthesis in Leishmania promastigotes. Instead, recent reports indicate that trypanosomes do synthesize PS via this pathway by head group Table 1. Overview of the observed retention times in the HPLC chromatogram shown in Figure 1 and their assignment to the individual L. donovani phospholipid species.  exchange with PE [43,44,45]. Given that all experiments undertaken in the present work were performed on the promastigote stage of the parasite, it might be possible that Leishmania amastigotes synthesize this lipid and regulate its transbilayer distribution in the parasite plasma membrane. Furthermore, loss of plasma membrane lipid asymmetry as a mechanism for survival in the host has been described for several protozoan parasites [46]. The analytical procedures described here should be useful for defining more precisely the phospholipid types exposed on their cell surfaces and help to uncover their potential role in parasite infectivity.

Materials
Annexin V-FITC and Annexin V-Alexa 569 were purchased from VPS-Diagnostics (Hoeven, The Netherlands) and Roche Diagnostics (Mannheim, Germany), respectively. PI was purchased as ammonium salt solution in chloroform from Sigma-Aldrich (Taufkirchen, Germany); all other lipids were obtained from Avanti Polar Lipids (Birmingham, AL, USA). High performance TLC silica gel 60 plates were from Merck (Darmstadt, Germany). Lipids were used without further purification. All solvents used were of the purest grade available. Unless indicated otherwise, all other chemicals and reagents were obtained from Sigma-Aldrich (Taufkirchen, Germany) and used as supplied.

Preparation of Giant Vesicles
Giant unilamellar vesicles were produced from lipid films dried on indium tin oxide (ITO) coated glass slides by electroswelling as originally described by Angelova et al. [47]. Shortly, lipid mixtures were made from 5 mM stock solutions in chloroform kept at 220uC. Lipids were mixed in a final volume of 50 mL chloroform (2 mM final total lipid concentration). This solution was dried under a gentle stream of N 2 to form a lipid film and then resolubilized in 80 mL chloroform. Single droplets of this lipid mixture were spread onto two ITO coated glass slides (Prä zisions Glas & Optik GmbH, Iserlohn, Germany). To obtain homogeneously distributed lipid films, the solvent was evaporated on a heater plate at 50-60uC. To get rid of traces of the solvent glass slides were placed under vacuum (,40 mbar) for 1 h. The electroswelling chamber was assembled from both lipid ITO coated slides using 1 mm teflon spacers. The chamber was filled with 1.5 ml of sucrose-buffer (250 mM sucrose; 7.5 mM NaN 3 ). An alternating voltage (rising from 0.02 V to 1.1 V over 30 min) with a frequency of 4 Hz was applied and the frequency was raised to 10 Hz in 30 s. Vesicles were formed during 3 h incubation. To detach the vesicles, a voltage of 1.3 V (4 Hz) was applied for 30 min. The vesicles were stored at room temperature for up to 5 days. For microscopy, the vesicle solution was diluted 1:2 with binding buffer (150 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM HEPES, pH 7.4) with addition of either 2 mM MgCl 2 (Ca 2+ -free binding buffer) or 2 mM CaCl 2 (Ca 2+ -containing binding buffer) .

In vitro Lipid Binding Assay with Annexin V-FITC
For protein lipid overlay assays, nitrocellulose membranes were first spotted with various indicated lipid species (500 pmol) dissolved in chloroform and blocked in binding buffer supplemented with 2% (w/v) fatty acid-free BSA for 30-60 min. After blocking, membranes were incubated for 15 min in the darkness with annexin V-FITC (250 ng/mL) in Ca 2+ -free or Ca 2+containing binding buffer. Thereafter, blots were washed three times in the respective binding buffer and scanned for FITC fluorescence using a Typhoon Trio variable-mode imager (GE Healthcare, Uppsala, Sweden) equipped with a 488 nm argon laser and a 526 nm short pass filter. To study annexin-V binding to lipid membranes, giant vesicles were diluted 1:2 with binding buffer supplemented with 0.2% (w/v) fatty acid-free BSA and 250 ng/mL annexin V-FITC. After 10 min incubation in the darkness, GUVs were either observed in differential interference contrast (DIC) or in fluorescence mode (excitation filter BP 470/40, beam splitter 500, band-pass filter 525/50) using an inverted fluorescence microscope (Leica DM 4000 B, Wetzlar, Germany). Images were acquired with a black-andwhite camera (Leica DFC340 FX). Fluorescent images were coloured using the green LUT settings of the program LAS AF Lite (Leica). Flow cytometry analysis was performed on a FACS Calibur instrument (BD Biosciences, San Jose, CA) equipped with an argon laser (488 nm) using Cell Quest software. Fluorescence was detected through a 530/30 band-pass filter. For each sample, data from 10,000 events were collected without gating. All experiments were performed at ambient conditions.

Lipid Extraction and Thin-layer Chromatography
Log phase promastigotes (10 8 cells) were harvested by centrifugation (10006g, 10 min), washed twice with PBS and suspended in PBS. Total cellular lipids were extracted by the method of Bligh and Dyer [48] and applied on TLC either manually or by means of the sample applicator Linomat IV (Camag, Muttenz, Switzerland). For one-dimensional TLC, plates were developed either in chloroform/ethanol/water/ triethylamine (30/35/7/35, v/v/v/v), in chloroform/metha- Table 2. Overview of the observed m/z values in the MALDI-TOF mass spectra shown in Figure 2 and their assignment to individual L. donovani phospholipid species.  nol/water (65/25/4, v/v/v), or chloroform/methanol/25% aqueous ammonium hydroxide (90/54/7, v/v/v). For twodimensional thin-layer chromatography, plates were first developed in chloroform/methanol/25% aqueous ammonium hydroxide (90/54/7, v/v/v) followed by chloroform/acetone/ methanol/acetic acid/water (50/20/10/10/5, v/v/v/v/v) for the second dimension. Lipids and standards were visualized with common lipid-locating agents such as iodine, ninhydrin (0.25% ninhydrin w/v in acetone), or primuline (5 mg primuline in 100 ml aceton/water, 80/20, v/v). The spots were assessed using a digital image system in combination with the program Argus X1 (BioStep, Jahnsdorf, Germany) and a Typhoon Trio variable-mode imager (GE Healthcare). For further analysis, lipid spots from primuline stained 1D-TLC or iodine stained 2D-TCL were scraped off and extracted as described for parasites. Lipids (extracted from TLC plates after separation under alkaline conditions) had to be protonated to regain binding affinity of annexin V-FITC. For this, recovered lipids were solubilized in chloroform/methanol (2:1, v/v) at a concentration of about 0.4 mg/ml. Next, one fifth of the solutions volume was added as water to obtain an aqueous layer and to enable phase separation. Thereafter, 1 mmol HCl was added and the solution was mixed. After centrifugation (5006g, 1 min), the chloroform phase was collected, dried and used for lipid protein overlay assays [27]. The 3 H-containing radioactive lipids were detected by fluorography; TLC plates were dipped in 0.4% 2,5-diphenyloxazol dissolved in 2-methylnaphthalene supplemented with 10% xylene and exposed to Kodak X-Omat S films at-80uC. Total phospholipid content was determined by digesting lipid samples in deionized water and perchloric acid for 1 h at 180uC followed by addition of ammonium molybdate and ascorbic acid. After further heating for 5 min in a boiling water bath, the sample was cooled and the absorbance was read at 797 nm to quantify total lipid phosphorus.